Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A computer-implemented method comprising: learning a probability distribution of a manufacturing system's performance conditioned on a training dataset comprising a historical tensor and associated performance metric of a reference period; receiving an input tensor associated with a time window and the input tensor's associated performance metric, the input tensor comprising at least multiple sensor variables associated with the manufacturing system and multiple steps of the manufacturing system's manufacturing process; based on the probability distribution, determining an overall change between the training dataset's relationship of the historical tensor and associated performance metric, and the relationship of the input tensor and the input tensor's associated performance metric; based on the probability distribution, determining a contribution of at least one of the multiple variables and the multiple steps to the overall change; and automatically triggering an action in the manufacturing system which reduces the overall change, wherein the determining of the contribution comprises implementing a Kullback-Leibler divergence between the probability distribution of the training dataset and the input tensor associated with the time window and the input tensor's associated performance metric.
Manufacturing process monitoring and control. This invention addresses the problem of identifying and mitigating performance degradation in manufacturing systems. It involves a computer-implemented method that first learns a probability distribution representing the historical performance of a manufacturing system. This learning is based on a training dataset containing past tensor data (representing system states) and corresponding performance metrics from a reference period. The method then receives new input tensor data, also associated with a performance metric, for a specific time window. This input tensor includes multiple sensor readings and information about different stages of the manufacturing process. Using the learned probability distribution, the system calculates an overall change in the relationship between system state and performance, comparing the historical data to the current input data. Furthermore, the method determines how much each individual sensor variable and each manufacturing step contributes to this overall change. This contribution analysis is performed by calculating the Kullback-Leibler divergence between the probability distribution derived from the training data and the probability distribution of the input tensor. Based on this analysis, an action is automatically initiated within the manufacturing system to reduce the identified overall change, thereby improving performance.
2. The method of claim 1 , wherein the automatically triggering the action comprises triggering a maintenance action associated with the at least one of the multiple variables and the multiple steps which contributed to the overall change.
This invention relates to automated maintenance systems for industrial processes or machinery, addressing the challenge of identifying and responding to operational changes that may require corrective action. The system monitors multiple variables and steps within a process, analyzing their contributions to an overall change in system behavior. When a significant change is detected, the system automatically triggers a maintenance action specifically linked to the variables or steps that contributed most to the change. This targeted approach improves efficiency by focusing maintenance efforts on the root causes of deviations, rather than performing broad or unnecessary interventions. The system may also prioritize actions based on the severity or impact of the contributing factors, ensuring timely and effective maintenance. By automating this process, the invention reduces manual oversight requirements and minimizes downtime, enhancing overall system reliability and performance. The method integrates real-time data analysis with predefined maintenance protocols, allowing for adaptive responses to dynamic operational conditions.
3. The method of claim 1 , wherein the automatically triggering an action comprises: searching a database comprising fault entries, each fault entry comprising a fault identifier, a reference signature and a maintenance action, for a match between the reference signature and a signature associated with the at least one of the multiple variables and the multiple steps which contributed to the overall change; and automatically triggering the maintenance action associated with the reference signature that matched.
This invention relates to automated fault detection and maintenance in industrial systems. The system monitors multiple variables and steps in a process to detect deviations or changes that may indicate faults. When a significant change is detected, the system automatically triggers a maintenance action by searching a database of fault entries. Each fault entry contains a fault identifier, a reference signature, and a corresponding maintenance action. The system compares the detected change's signature against the reference signatures in the database. If a match is found, the associated maintenance action is automatically executed. This approach ensures timely and accurate fault resolution by leveraging pre-defined fault signatures and actions, reducing manual intervention and improving system reliability. The method enhances predictive maintenance by correlating process variables and steps to known fault patterns, enabling proactive maintenance before failures occur. The database-driven approach allows for easy updates and scalability, accommodating new fault signatures and maintenance procedures as needed.
4. The method of claim 1 , wherein the automatically triggering an action comprises: automatically executing a generic maintenance action.
A system and method for automated maintenance in industrial or mechanical systems addresses the problem of inefficient and reactive maintenance processes, which can lead to downtime and increased costs. The invention provides a solution by automatically detecting operational anomalies or predefined conditions in a system and triggering maintenance actions without manual intervention. The method involves monitoring system parameters, such as performance metrics or sensor data, to identify deviations from normal operation. When an anomaly is detected, the system automatically executes a generic maintenance action, such as a diagnostic check, calibration, or component replacement, to restore optimal functionality. The maintenance actions are predefined and standardized, ensuring consistency and reducing the need for human decision-making. This approach improves system reliability, minimizes unplanned downtime, and lowers maintenance costs by proactively addressing potential issues before they escalate. The system may also log maintenance events for future reference and analysis, enabling continuous improvement of maintenance protocols. The invention is particularly useful in industries where continuous operation is critical, such as manufacturing, energy production, and transportation.
5. The method of claim 1 , wherein the method is implemented in an automated online maintenance tool monitoring the manufacturing system in real-time.
This invention relates to automated online maintenance tools for real-time monitoring of manufacturing systems. The technology addresses the challenge of efficiently detecting and diagnosing maintenance issues in industrial equipment to minimize downtime and improve operational efficiency. The system continuously monitors the manufacturing system in real-time, collecting data from various sensors and components to identify potential failures or performance deviations. The automated tool processes this data using predefined algorithms and machine learning models to detect anomalies, predict maintenance needs, and generate alerts for corrective actions. The tool also provides diagnostic insights, suggesting specific maintenance procedures or component replacements based on the analyzed data. By integrating real-time monitoring with predictive analytics, the system enables proactive maintenance, reducing unplanned downtime and extending the lifespan of manufacturing equipment. The automated nature of the tool ensures continuous, unbiased monitoring, improving reliability and operational efficiency in industrial environments.
6. The method of claim 1 , wherein the action is automatically triggered responsive to the overall change meeting a threshold value.
A system and method for automated action triggering based on detected changes in a monitored environment. The invention addresses the problem of manually monitoring and responding to environmental changes, which is inefficient and prone to human error. The system continuously monitors one or more environmental parameters, such as temperature, humidity, or motion, and calculates an overall change in these parameters over time. When the overall change meets or exceeds a predefined threshold value, the system automatically triggers a predefined action, such as sending an alert, activating a device, or adjusting system settings. The threshold value can be dynamically adjusted based on historical data or user preferences. The system may also include filtering mechanisms to reduce false positives, such as ignoring transient changes or requiring confirmation from multiple sensors. The invention improves efficiency and reliability in automated monitoring and response systems by eliminating the need for manual intervention.
7. A system comprising: at least one hardware processor coupled with a manufacturing controller receiving sensor data associated with a manufacturing system; a memory coupled with the at least one hardware processor; the at least one hardware processor operable to at least: learn a probability distribution of the manufacturing system's performance conditioned on a training dataset comprising a historical tensor and associated performance metric of a reference period; receive an input tensor associated with a time window and the input tensor's associated performance metric, the input tensor comprising at least multiple sensor variables associated with the manufacturing system and multiple steps of the manufacturing system's manufacturing process; based on the probability distribution, determine an overall change between the training dataset's relationship of the historical tensor and associated performance metric, and the relationship of the input tensor and the input tensor's associated performance metric; based on the probability distribution, determine a contribution of at least one of the multiple variables and the multiple steps to the overall change; and automatically trigger an action in the manufacturing system which reduces the overall change, wherein the hardware processor determines the contribution by implementing a Kullback-Leibler divergence between the probability distribution of the training dataset and the input tensor associated with the time window and the input tensor's associated performance metric.
This invention relates to a system for monitoring and optimizing manufacturing processes using machine learning. The system addresses the challenge of identifying and mitigating performance deviations in manufacturing operations by analyzing sensor data and process steps to determine their impact on overall performance. The system includes a hardware processor coupled with a manufacturing controller that receives sensor data from a manufacturing system. The processor learns a probability distribution of the system's performance based on historical data, including sensor variables and process steps from a reference period. It then receives real-time input data from a time window, including sensor variables and process steps, along with an associated performance metric. Using the learned probability distribution, the system calculates the overall change in performance between the historical reference data and the current input data. It further determines the contribution of individual sensor variables and process steps to this performance change. The system employs Kullback-Leibler divergence to quantify the difference between the probability distributions of the reference and current data. Based on these calculations, the system automatically triggers corrective actions in the manufacturing system to reduce performance deviations. This approach enables real-time process optimization by identifying and addressing specific factors influencing performance.
8. The system of claim 7 , wherein the hardware processor automatically triggers the action comprising a maintenance action associated with the at least one of the multiple variables and the multiple steps which contributed to the overall change.
This invention relates to a system for automated maintenance in industrial or operational environments. The system monitors multiple variables and steps in a process to detect changes that may indicate potential issues. When a significant change is detected, the system automatically triggers a maintenance action. The maintenance action is specifically associated with the variables or steps that contributed to the detected change, ensuring targeted and efficient intervention. The system uses a hardware processor to analyze data from sensors or other monitoring devices, identify deviations from expected behavior, and determine the appropriate maintenance response. This approach reduces downtime and improves operational efficiency by addressing issues before they escalate. The system may be applied in manufacturing, energy production, or other industries where continuous monitoring and predictive maintenance are critical. The invention focuses on automating the maintenance process to minimize human intervention and enhance reliability.
9. The system of claim 7 , wherein the hardware processor automatically triggers the action by searching a database comprising fault entries, each fault entry comprising a fault identifier, a reference signature and a maintenance action, for a match between the reference signature and a signature associated with the at least one of the multiple variables and the multiple steps which contributed to the overall change, the hardware processor automatically triggering the maintenance action associated with the reference signature that matched.
The system is designed for automated fault detection and maintenance in industrial or technical systems. The problem addressed is the need for real-time identification of system faults and the automatic execution of corrective actions without manual intervention. The system monitors multiple variables and steps within a process, tracking changes in these parameters to detect anomalies or deviations from expected behavior. When a significant change is detected, the system generates a signature representing the contributing variables and steps. This signature is then compared against a database of fault entries, where each entry includes a fault identifier, a reference signature, and a predefined maintenance action. The system automatically searches the database for a match between the generated signature and the reference signatures stored in the database. If a match is found, the associated maintenance action is triggered automatically, ensuring rapid response to detected faults. This approach reduces downtime and improves system reliability by leveraging preconfigured fault signatures and corresponding corrective measures. The system is particularly useful in environments where continuous monitoring and immediate fault resolution are critical, such as manufacturing, energy production, or automated control systems.
10. The system of claim 7 , wherein the hardware processor automatically triggers the action by automatically executing a generic maintenance action.
11. The system of claim 7 , wherein the hardware processor implements an automated online maintenance tool monitoring the manufacturing system in real-time.
The system relates to automated monitoring and maintenance of manufacturing systems. Traditional manufacturing systems often require manual inspections and maintenance, leading to inefficiencies, unplanned downtime, and increased operational costs. This system addresses these issues by integrating an automated online maintenance tool that continuously monitors the manufacturing system in real-time. The tool uses a hardware processor to analyze data from sensors, control systems, and other components within the manufacturing environment. By detecting anomalies, performance deviations, or potential failures early, the system enables proactive maintenance, reducing downtime and improving overall system reliability. The maintenance tool may also generate alerts, suggest corrective actions, or trigger automated responses to mitigate issues before they escalate. This real-time monitoring capability enhances operational efficiency, minimizes human intervention, and extends the lifespan of manufacturing equipment. The system is particularly useful in industrial settings where continuous operation and minimal disruptions are critical.
12. The system of claim 7 , wherein the action is automatically triggered responsive to the overall change meeting a threshold value.
A system for monitoring and responding to changes in a physical or digital environment detects variations in conditions such as temperature, pressure, or data values. The system includes sensors or data sources that measure these conditions and a processing unit that calculates an overall change in the monitored parameters over time. When the overall change exceeds a predefined threshold value, the system automatically triggers a predefined action. This action may include generating an alert, adjusting a control mechanism, or initiating a corrective procedure. The system may also include a user interface for configuring the threshold value and the specific action to be taken. The processing unit continuously evaluates the monitored data to determine whether the threshold is met, ensuring timely responses to significant changes. This automated triggering mechanism reduces the need for manual intervention and improves system efficiency by responding to deviations in real time. The system is applicable in industrial automation, environmental monitoring, or data management scenarios where rapid detection and response to changes are critical.
13. A computer program product comprising a computer readable storage medium having program instructions embodied therewith, wherein the computer readable storage medium is not a transitory signal per se, the program instructions readable by a processor to cause the processor to perform a method comprising: learning a probability distribution of a manufacturing system's performance conditioned on a training dataset comprising a historical tensor and associated performance metric of a reference period; receiving an input tensor associated with a time window and the input tensor's associated performance metric, the input tensor comprising at least multiple sensor variables associated with the manufacturing system and multiple steps of the manufacturing system's manufacturing process; based on the probability distribution, determining an overall change between the training dataset's relationship of the historical tensor and associated performance metric, and the relationship of the input tensor and the input tensor's associated performance metric; based on the probability distribution, determining a contribution of at least one of the multiple variables and the multiple steps to the overall change; and automatically triggering an action in the manufacturing system which reduces the overall change, wherein the determining of the contribution comprises implementing a Kullback-Leibler divergence between the probability distribution of the training dataset and the input tensor associated with the time window and the input tensor's associated performance metric.
This invention relates to predictive analytics in manufacturing systems, specifically for identifying and mitigating performance deviations. The system learns a probability distribution of a manufacturing system's performance using historical data, including sensor variables and process steps, along with associated performance metrics from a reference period. When new input data is received, the system compares it to the learned distribution to determine overall performance changes and the specific contributions of individual variables or process steps to those changes. The comparison uses Kullback-Leibler divergence to quantify differences between the historical and current data distributions. Based on this analysis, the system automatically triggers corrective actions to reduce performance deviations. The method enables real-time monitoring and optimization of manufacturing processes by isolating root causes of inefficiencies and applying targeted interventions. The approach leverages machine learning to model complex relationships between process variables and performance outcomes, improving operational efficiency and product quality.
14. The computer readable storage medium of claim 13 , wherein the automatically triggering the action comprises triggering a maintenance action associated with the at least one of the multiple variables and the multiple steps which contributed to the overall change.
This invention relates to predictive maintenance systems that analyze operational data to identify and address potential issues in industrial or mechanical systems. The problem addressed is the need for automated, data-driven maintenance actions that prevent failures or inefficiencies by detecting subtle changes in system variables and processes. The system monitors multiple variables and steps within a process, tracking their individual and collective contributions to overall system performance. When a significant change in performance is detected, the system automatically triggers a maintenance action. This action is specifically tailored to the variables or steps that contributed most to the observed change, ensuring targeted and efficient intervention. The system may use historical data, machine learning, or statistical analysis to determine the root causes of performance deviations and recommend or execute corrective measures. The invention improves upon traditional maintenance approaches by reducing downtime, minimizing unnecessary interventions, and optimizing resource use. By focusing on the specific variables or steps that caused the performance change, the system avoids broad, inefficient maintenance actions and instead applies precise solutions. This approach is particularly useful in complex systems where multiple interdependent factors influence performance, such as manufacturing equipment, industrial machinery, or automated production lines. The system may integrate with existing monitoring tools or standalone sensors to collect real-time data, ensuring timely and accurate maintenance decisions.
15. The computer readable storage medium of claim 13 , wherein the automatically triggering an action comprises: searching a database comprising fault entries, each fault entry comprising a fault identifier, a reference signature and a maintenance action, for a match between the reference signature and a signature associated with the at least one of the multiple variables and the multiple steps which contributed to the overall change; and automatically triggering the maintenance action associated with the reference signature that matched.
This invention relates to automated fault detection and maintenance in industrial or technical systems. The problem addressed is the need for efficient and accurate identification of system faults and the automatic execution of corrective actions without manual intervention. The system monitors multiple variables and steps within a process, detecting changes that indicate potential faults. When a significant change is detected, the system generates a signature representing the variables and steps contributing to the change. This signature is compared against a database of fault entries, each containing a fault identifier, a reference signature, and a predefined maintenance action. If a match is found between the generated signature and a reference signature, the corresponding maintenance action is automatically triggered. The database allows for rapid fault identification and response, reducing downtime and improving system reliability. The invention ensures that maintenance actions are based on precise, data-driven matches, minimizing errors and unnecessary interventions. The automated process enhances efficiency by eliminating the need for manual fault diagnosis and decision-making.
16. The computer readable storage medium of claim 13 , wherein the automatically triggering an action comprises: automatically executing a generic maintenance action.
A system and method for automated maintenance in computing environments addresses the challenge of efficiently managing routine maintenance tasks without manual intervention. The invention involves a computer-readable storage medium containing instructions that, when executed, perform automated maintenance actions based on predefined conditions. The system monitors system parameters, such as performance metrics, error logs, or resource utilization, to detect when maintenance is required. Upon detecting a trigger condition, the system automatically executes a generic maintenance action, such as software updates, disk cleanup, or system reboots, to ensure optimal performance and reliability. The maintenance actions are predefined and standardized, ensuring consistency and reducing the need for manual oversight. This approach minimizes downtime, improves system efficiency, and reduces the risk of human error in maintenance operations. The system may also log maintenance activities for auditing and reporting purposes. The invention is particularly useful in large-scale computing environments where manual maintenance is impractical or inefficient.
17. The computer readable storage medium of claim 13 , wherein the action is automatically triggered responsive to the overall change meeting a threshold value.
This invention relates to a system for monitoring and analyzing changes in data, particularly for detecting significant variations that may require automated intervention. The system involves a computer-readable storage medium containing instructions that, when executed, perform a method of tracking changes in a dataset over time. The method includes calculating an overall change value representing the cumulative difference between current and previous data states. The system then compares this overall change value against a predefined threshold to determine whether the change is significant enough to warrant action. If the threshold is met or exceeded, the system automatically triggers a predefined action, such as generating an alert, updating a database, or initiating a corrective process. The threshold can be dynamically adjusted based on historical data patterns or user-defined criteria to ensure accurate detection of meaningful changes. This approach ensures timely responses to critical data variations without manual intervention, improving efficiency and reliability in data management systems. The invention is particularly useful in applications where real-time monitoring and automated responses to data fluctuations are essential, such as financial systems, network security, or industrial process control.
Unknown
August 25, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.